Learning Structure-Thermal Property Relationships in 2D Materials*

Two dimensional monolayer semiconductors, alloys and patterned lateral heterostructures are extremely promising candidates for the next generation of nanoelectronic devices. Quantification of thermal transport of such two dimensional materials and heterostructures is necessary for the design of such nanoelectronic and thermoelectric devices. However, direct experimental measurements of intrinsic thermal conductivity is challenging at these length scales and, therefore, the role of material stoichiometry and phase distribution on thermal transport properties of these materials remains unknown.
Here, we use fully atomistic non-equilibrium molecular dynamics simulations to compare the calculated intrinsic thermal conductivity of a Mo_1-xW_xSe₂ monolayer alloy with that of a self-similar fractal MoSe₂/WSe₂ heterostructure. Machine learning applied to the compositional phase space of these materials is used to predict heterostructures with desired thermal transport properties.